From use case to an AI-enabled process in production, integrated with the company's systems
We take into production a use case that has already been chosen, through a Mapping or by the company itself, and we consider the work finished when the team that runs the process uses it every day, with a measure of what has changed.
Solution 02 of 04
How do you get to production?
AI-enabled processes in production
From defining the case to release into production
Our forward deployed engineers, specialists in data and AI, work on site at the company alongside the people who know the process, because many rules only emerge by watching the work as it happens. They stay with the client's team from the analysis of requirements and the preparation of the data, often scattered across several systems, through to release into production and full adoption of the solution, without stopping at a prototype.
Definition
With the process owner we set the objective, the baseline and the acceptance criteria
Preparation
We connect the sources, check the data, and configure access and permissions
Build
We build the agents and write the rules on what they may do
Testing
The agents work on real cases with known outcomes; we release only above the agreed thresholds
Release
We train the team, the process goes into production and measurement begins
The modules, activated according to what the process needs
For each process we activate the modules it needs and, where it makes sense, combine them with third-party technology and bespoke development; this is how the company's Data & AI platform takes shape.
Data Foundation
The process data, with a definition, an owner and a quality rule
Data Intelligence
The analysis and scoring models, for example to prioritise cases
Company Brain
The written rules, the exceptions, the approval criteria
Agents
The process agents and their test bench
Workspace
The approval queue and the process screens
Governance
The register of agents, models and runs, with the evidence
Security
Agent identities and permissions, data protection
How much the agent does and how much the person decides
Companies usually start at the first or second level: in a claims office the agent prepares the settlement proposal and the claims handler decides. Irreversible decisions and those that affect the end customer always stay with a person.
Agents work at three levels of autonomy, from observation, to preparing a complete piece of work, to carrying out defined actions within agreed thresholds. The move from one level to the next is decided by the process owner on the basis of the test results, as we describe on the Agents module page.
The same measure before and after release
Before building, we measure the process's handling time, errors and rework; after release we record the same things, together with the time spent checking the agents. The time freed up becomes value when the company decides how to use it.
Frequently asked questions
Is the result guaranteed?
We commit to releasing a solution that has passed the agreed tests, while the economic result is something we measure with you, on your numbers.
What if the tests are not passed?
Then we do not release. The thresholds are agreed before building, and the test report explains why the project stops.
Who accepts the solution before release?
The process owner accepts the measure and the criteria; IT and security, the integrations and permissions; risk and compliance, the tests and thresholds.
Will we end up depending on you?
The rules, data and documentation of the process remain yours, and the knowledge transfer is written into the contract. Presidio is optional.
Assessing a use case
In the first conversation we look with you at the process, the data and the systems involved, to understand whether the project will take a few weeks or a few months.